Sense solar setup instructions seem wrong

Technical Questions
21

I think what I’ll do is just swap my CTs back to where they were and wait until my new CTs arrive before trying recalibration. When I do that I’ll pay attention to which lead in the app is dropping off to make sure they’re on the correct phase.

The more I think about this - the more it could be contributing to my production errors. If things are being approved manually and the software is assuming I have my CTs on the right leg - when it’s doing its calculations it may be subtracting from one leg when it should be subtracting from the other, and vice versa.

22

Thats it - when you go into the settings - > sense monitor -> Solar (enabled), by clicking on Solar you’ll be walked through the calibration process again.

23

[quote=“NJHaley, post:7, topic:380, full:true”]
You’re the second member I’ve seen reporting that here, you’ll have to keep trying to get hold of tech support, there’s something they should be able to do to get you back on track.[/quote]

I’ve repeatedly tried to get hold of tech support, and they just tell me to log out of the iPhone app and log back in again. Of course, that does nothing to help.

I’ve been very disappointed, but to be fair, it’s not really their skill set. The rôle of tech support is for when you have a mature product that works, and the customer is doing something wrong, and the job of tech support is to explain to the customer what they’re doing wrong.

The purpose of a beta test program is for a product that doesn’t quite work yet, and you enlist the help of experienced, articulate people, who can help the engineers work out why it isn’t working, so they can fix it.

Sense seems to have tech support staff taking the place of the engineers in the beta test program, which isn’t a lot of use because the tech support staff can’t actually do anything to fix the problems reported by the beta testers.

So we just end up with tech support saying, “Try turning it off and on again,” because there’s nothing else they can do.

1 Like
24

[quote=“NJHaley, post:16, topic:380”]
My impression was that the labels were meant to be facing wherever the power was coming from, solar or utility, but you’re right, the way the instructions read it’s hard (for me at least) to tell - since the breaker I used to shut off power during calibration was between my panel and my inverter. Obviously, I have another breaker that shuts off solar power right on my panel - is that the one that is meant to be shut off during calibration?[/quote]

It doesn’t matter. It’s a circuit. If any switch in the circuit is off, no current is flowing.

25

[quote=“miracj, post:10, topic:380”]
Since the power of the left and right sides of the solar generated power generally are close to identical, it may not matter a lot, but having them aligned with the mains seem right to me.[/quote]

It’s true that the difference between the two solar phase is tiny, but since the whole point of the Sense monitor is that it makes very precise measurements and detects tiny variations in the load, having constant random phantom fluctuations is likely to interfere with the algorithm. It seems that having constant random fluctuations of ±10W on each phase (at least during the day when the sun is shining) is going to make it much harder for the Sense monitor to detect a 10W light turing on or off.

26

[quote=“miracj, post:10, topic:380”]
However, once I swapped the solar CTs, the total of the mains is now the house power usage minus the solar power generated. Need to address that.[/quote]

When you swap the CTs, remember to flip the orientation of the label side too. (The CT is now on the opposite phase, so the current is going the opposite way at any point in the sine-wave cycle, so you need to turn the CT around to face the opposite direction too.)

27

I just hope they’re able to enlist and train more tech support - it’s obviously the weakest link in their lineup. Not because the support they have aren’t helpful, they try their best, but they’re overwhelmed. I have a feeling their sales have ramped up, especially after the TOH episode, and tech support is going to be stretched even thinner. Customers discouraged by the lack of tech support is easily the biggest complaint I’ve seen, it’s a turn off for prospective buyers and makes the customer feel like they got a lemon.

I’ve bought a lot of smarthome products - ecobee, smartthings, rachio among others - they all have community sites like this one and very prolific support personnel on those sites so the community feels like they’re valued and can get their problems solved quickly. Sense needs that presence here badly, in my opinion. It makes the consumer feel stiffed otherwise.

28

[quote=“miracj, post:19, topic:380”]
With Solar on and Solar off during a sunny day, both solar sides should be positive when on, and close to zero when off. If when Solar is on, either side was negative, the software could mark the direction of each Solar CT.[/quote]

This is alternating current (AC), not direct current (DC). Talking about positive and negative is not useful. Both lines are both positive and negative sixty times per second, at opposite times. Generally, at any instant, one line is positive when the other one is negative.

How you tell whether something is providing power or consuming it is by comparing the phase of the current sine wave with the phase of the voltage sine wave.

If, when the voltage is positive, the current is flowing out of the electrical panel, that means the device at the other end of that wire is receiving energy.

If, when the voltage is positive, the current is flowing into the electrical panel, that means the device at the other end of that wire is producing energy.

And vice versa, when the voltage is negative.

This is why, when the solar current sensors are installed the wrong way, with the label towards the panel instead of towards the energy source, the Sense monitor gets the phases backwards.

When the L1 voltage is positive and the L1 solar current is flowing into the electrical panel, if the L1 solar current sensor is backwards, then that looks like current flowing out of the electrical panel, which doesn’t make any sense, since we know solar panels are supposed to produce power, not consume it.

But when the L1 voltage is positive the L2 voltage is negative, which means that the L2 solar current is supposed to be flowing out of the electrical panel. So the Sense monitor concludes that the backwards current sensor on L1 is actually on L2, and gets the phases switched.

What shocks me about this is that the whole idea behind Sense is that it can deduce the subtlest nuances of your home electrical setup through Machine Learning and Artificial Intelligence. And yet somehow it hasn’t spotted that they’re telling everyone to install their solar current sensors backwards.

29

Makes sense. I’m grasping at straws trying to figure out why my solar is underreporting by ~3.75% regularly. When it’s dark, my consumption is dead on. During the day, my consumption is adversely affected by the solar miscalculation. The most recent straw was replacing the cts, thinking there was maybe a production defect in the ones I have. When I read your post, I started wondering how that may affect my readings. Still unsure, but I’ll know more when the new cts arrive and I can do some more trial and error. Appreciate your post, it’s been helpful.

30

[quote=“Stuart, post:28, topic:380”]
When you swap the CTs, remember to flip the orientation of the label side too. (The CT is now on the opposite phase, so the current is going the opposite way at any point in the sine-wave cycle, so you need to turn the CT around to face the opposite direction too.)[/quote]
I don’t believe that is correct. The two labels are made to indicate which way to hook up the CTs, not their phase, so they are either opposite phase with the labels facing the same way, or they do it in software, changing them to be opposite.

[quote=“Stuart, post:28, topic:380”]
This is alternating current (AC), not direct current (DC). Talking about positive and negative is not useful. Both lines are both positive and negative sixty times per second, at opposite times. Generally, at any instant, one line is positive when the other one is negative.[/quote]
We’re talking negative wattage value displays, not voltage. Negative wattage displayed is just the power going the other way (Production vs consumption)

That is how you tell power in an inductive or capacitive load., not relevant here. Sense is displaying wattage for each phase for the mains and the solar. Because of their algorithm, they are displaying the numbers being used and that being produced, but don’t always compute the correct value based upon where the solar power is being fed into the system. E.g. My solar feeds directly into the utility mains. Thus the mains power needs to have effectively remove the power generated by the solar.In my most recent case, I have positive solar values and a negative mains value that is the value of house usage - the larger solar value for that side… It’s all software settings (values we can’t set in the app)

31

[quote="miracj, post:30, topic:380”]
The two labels are made to indicate which way to hook up the CTs, not their phase, so they are either opposite phase with the labels facing the same way, or they do it in software, changing them to be opposite.

Sense is displaying wattage[/quote]
Unfortunately CTs (current transducers) don’t measure wattage (power), they measure current* (hence the name). To calculate power you have to multiple instantaneous current by instantaneous voltage. And since there are two instantaneous voltages on the two phases (with opposite sign), the Sense monitor has to pick the right one to multiply by, or it gets a negative value. It picks the voltage that yields a positive power, since it knows solar panels are supposed to generate power, not consume it. But if the current transducer is installed backwards, the current it reads will be the negative of the correct current, causing the Sense monitor to pick the other phase voltage to multiply by to get a positive value, causing the Sense monitor to associate that current transducer with the wrong phase.

The convention is that current transducers generally have the label on the side towards the power supply. By installing them as the instructions say with label on the side towards the load, the sign of the current is reversed, and the Sense monitor associates that current transducer with the opposite phase.

*Actually, to be precise, a current transformer produces a voltage proportional to the change in current, which is why they don’t work for DC, only AC (where the current is constantly changing). The rate of change is highest at the zero crossing, so the peak CT voltage actually occurs at zero current. If there’s a constant DC bias on the line, a current transformer can’t detect that, but for household AC it’s safe to assume that the average DC current is zero (the positive and negative halves of the sine wave add up to zero). Since we can assume the average current is zero, from tracking the current change reported by the CT, we can infer the actual current flowing (and its direction) at any moment in time.

32

[quote=“NJHaley, post:29, topic:380, full:true”]
Makes sense. I’m grasping at straws trying to figure out why my solar is underreporting by ~3.75% regularly. When it’s dark, my consumption is dead on. During the day, my consumption is adversely affected by the solar miscalculation. The most recent straw was replacing the cts, thinking there was maybe a production defect in the ones I have. When I read your post, I started wondering how that may affect my readings. Still unsure, but I’ll know more when the new cts arrive and I can do some more trial and error. Appreciate your post, it’s been helpful.[/quote]

Having the solar current transducers associated with the wrong phase may affect the Sense monitor’s ability to detect small device loads reliably, but should not significantly affect the overall total power figures.

If your Sense monitor is consistently reporting 3.75% less power than your solar inverter is claiming, then perhaps your solar inverter is being overly generous in the power it is claiming to produce, or the Sense monitor is inaccurate, or maybe both are accurate and there’s 3.75% loss in the cables connecting your inverter to your main panel, although that seems like a lot. For a 10kW solar system 3.75% would be 375W lost as heat in the cables, which would make them quite warm. An infrared imager like a FLIR ONE would let you see if the cables are getting warm.

33

It’s even a little more complicated than that - it’s about 6.5% lower than the inverter is claiming, 3.75% lower than 2 meters are claiming - my solar meter (which measures how much I produce) and my utility meter (which measures how much I sell back to my utility).

Some things I’ve mathematically figured out: my inverter is obviously being generous, so I strictly rely on the meters. Second, the fraction Sense is underreporting as produced accounts entirely for the numerical underestimate of kwh used during the day, if that makes sense. So if we produce 40kwh (@meter), sense says we’ve made 38.5 (a difference of 1.5, or a loss of 3.75%). If, during the day, the meter says we used 10kwh, sense says we used 8.5 (again, a 1.5 kwh difference, but a loss of 15%!). At night, however, if the meter says we’ve used 10kwh, sense says we’ve used 10kwh - the power in this case is coming entirely from the grid, taking the solar side out of the equation. I’ve got a spreadsheet in another thread somewhere with about a week’s worth of data to show my work.

I have to think that might make learning very difficult, since the use values for appliances during the day are significantly different than those sense calculates at night (my baseline use during the day, for example can be as low as 30w, meanwhile at night it runs about 150w). Since tech support says there’s nothing wrong with the equations and it seems to be calculating correctly at night, one potential variable is the cts doing the measuring. A manufacturing defect would potentially keep the cts from measuring correctly (if they’re imperceptibly, incompletely closed for example). Alternatively, if I throw new cts on there and get the exact same results, the errors do lay on the software side of things.

Sorry for hijacking your thread, but thanks for letting me bend your ear.

34

[quote=“NJHaley, post:33, topic:380”]
Since tech support says there’s nothing wrong with the equations and it seems to be calculating correctly at night, one potential variable is the CTs doing the measuring. A manufacturing defect would potentially keep the CTs from measuring correctly (if they’re imperceptibly, incompletely closed for example). Alternatively, if I throw new CTs on there and get the exact same results, the errors do lay on the software side of things.[/quote]

It does sound like your solar CTs are under-measuring the current. It’s odd though that they’re under-measuring by a fixed amount, not a fixed percentage.

Let us know how you fare with the new CTs when they arrive.

35

Aha - but it is a fixed percentage with regard to solar. The way I knew it was solar and the solar cts alone was that the numerical value for that fraction was the exact numerical amount the usage was underreporting during the day. Had it been an across the board percentage, then you could argue the meters (or both sides of the sense calcs) were off. Another example - today we’re heading towards 15kwh production (overcast). I predict sense will say I produced 15x0.965 = 14.48. If my utility meter says we used 12kwh during the day, I’d predict sense would say we used 12-0.52kwh =11.48kwh.

Hard to follow, I know, it gave me a headache trying to figure it out, but one way I could test if it was the solar CT would be to physically swap cts. Unplug my solars and plug them into my usage plug, and vice versa. Connect them to the appropriate leads, etc. I’d expect then that my solar would read perfect, but my utility side would underreport. The math is probably a little more complicated than that :confused:

Checked my setup today, everything seems to be attached to the correct leads. Labels facing the inverter on the solar cts…

36

[quote=“NJHaley, post:35, topic:380”]
Hard to follow, I know, it gave me a headache trying to figure it out, but one way I could test if it was the solar CT would be to physically swap cts. Unplug my solars and plug them into my usage plug, and vice versa.[/quote]

If you can afford one, buy yourself a current clamp meter and use it to measure the solar lines. It will only measure current, not power, but since the solar voltage phase ought to be almost perfectly aligned with the current phase, you can just multiply by RMS voltage to get the power.

37

I just might. New CTs arrived today and will be installed this afternoon when I get home from work. My ADHD can’t bear to have this sort of data I’ll post below. 10w of usage sounds impressive, but our “Always on” is coming and going because during the day our apparent usage is dropping below 20-30w (sometimes not even registering!). Obviously we’re using more than 10w, but because our solar CTs aren’t reading correctly, Sense isn’t properly accounting for all of our usage - just a fraction of it (mainly what’s trickling in from the main lines, which are being read correctly).

I think these problems reported elsewhere are related:

Here are some readings from today:
Low readings

1440×2560 82.9 KB

No readings!!

1440×2560 80.9 KB

38

Yea there is something wrong as there is no way any house these days can only be using 10 watts of power. I think the instant on function on my three TVs uses more than that. Then add in your cable modem and your router, a few AC/DC converters, maybe an electric clock or two and right there you are well over 10 watts.

39

My baseline at night (main CTs only) is giving me reliable data, strongly implying either a defect at the level of the solar CTs or the calculations for solar power. We’ll find out today which it is. I strongly suspect it’s at the level of calculations - but I’m not getting any help from the Sense engineers on this.

Screenshot_20170221-131442.png1440×2560 73.5 KB

(#)firstworldproblems

40

New CTs, no change - the problem is inherent to the solar calculation algorithms and that’s as far as I care to do the work for them. I’m losing interest in a product with support and engineers who can’t even be bothered to respond to early adopter problems or constructive criticisms. This community site is a real letdown for its users.