New 3.5mm cables now utility goes negative when solar isn't generating?

I recently bought a few (24 to be exact) 3.5mm extender cables: https://www.amazon.com/gp/product/B086WSZHPB/ref=ppx_yo_dt_b_asin_title_o04_s00?ie=UTF8&psc=1

After powering down the IotaWatt and installing the new cables everything works. But now my Utility Power readings go negative when solar is NOT generating. When solar is generating everything is positive. Before Utility went negative when solar was generating. Which seemed correct as I was able to compute consumption with that situation. (UTILITY + SOLAR)ABS

Do the type of cables matter? As I bought a number of these as well to extend the length and these seemed to work ok:

Actually I noticed a difference the cables that work have three sections on the plug (I believe this is called TRS). The ones that seem to not be working have four sections on the plug (TRRS). Would that be causing these issues with the CT inputs?

IoTaWatt inputs will accept a mono (sleeve and tip) or stereo (sleeve, ring and tip) 3.5mm jack. I have never tried to use 4 conductor extensions, so can’t say what might be happening.

I would expect the mains to go negative with solar production, but not branch circuits. Maybe you could post some screenshots to illustrate the problem.

You also might try switching which CTs use the three and four conductor extensions to see if the problem moves.

@overeasy Attached are some screenshots. The Sense screenshot is for reference taken at the same time the screenshot of the IotaWatt’s status page was taken. Notice were generating 18,900 watts of solar and consuming 3,900. But Sense shows the utility (mains) still as positive. Also the only inputs on the IotaWatt that are configured to allow negative are the utility (mains) CT’s.

Sadly I don’t have any spare 3.5mm extensions right now. I ordered a few new TRS style extensions off amazon but they wont be in till Friday. :frowning:




Here is a Graph+ view starting from this past Monday 9/7.

I did test one of the TRRS cables out on the UtilityMain_PhaseA on Monday night (highlighted circle) and it seemed to be working as expected. But then when I unplugged everything and re-organized it for my project box last night (Tuesday Night 9/8) both UtilityMain’s are now not reporting correctly (highlighted rectangle). Nothing changed with the CT clamps and all the cables are labeled and double checked that they were plugged into the correct spots.

So I’m not really sure whats happening here. My best guess is TRRS vs TRS but what debunks that for me is when I plugged in the cable to test on phaseA it seemed to be reporting fine on Monday night.

I’m not seeing a negative input except the TeslaSubPanel_PhaseA.

I do see other issues that may need attention:

You seem to have CTs on both legs of the solar inverter output. Most inverters are two-wire and output the same on both legs. The IoTaWatt is reporting significantly different power on the two legs, which should not be the case. I believe the Sense only uses one CT for the solar. Is that true in your case?

A 20,000Watt solar array would put out more than 80Amps on each leg. I’m wondering how that is fed into your system. There are two different approaches:

Type 2 as it is commonly referred to, brings the inverter power into your panel through a circuit breaker at the bottom (opposite end from the mains). In your case that would probably be a 100A breaker or greater.

Type 1, as commonly referred, brings the inverter power through a breaker in a separate enclosure and connects the output to your mains before (upstream) from the mains.

With type 2, the inverter output reduces the mains load and drives the mains negative when the inverter outputs more power than you are using. Consumption is the algebraic sum of the mains and solar.

With type 1, the inverter output has no effect on the mains power reading, which is your consumption.

From this data, I believe you have type 2.

Moving on from that, I think the Sense and IoTaWatt agree within 1% of the Mains power.

I think the 3,983 that Sense is reporting is computed by adding a negative mains power of 14,936Watts to the positive Inverter output of 18,919Watts.

The IoTaWatt reports a similar number for the mains at 15,106, although it’s positive. I suggest that you should check “reverse” for UtilityMain_PhaseA and uncheck "reverse for UtilityMain_PhaseB. Then the mains should show negative and your Consumption output should be closer to the Sense.

If your inverter is truly two-wire, i.e. it has no third neutral conductor, then you should use just one CT and check the “double” box. But before you do that, could you move the CT that’s on phaseB to phase A and see if the two read differently while solar is producing. If they are different, could you try moving one of the Tesla CTs to the same wire and see how it agrees with the other two? There is something wrong here and this experiment might shed some light on it.

I don’t think the IoTaWatt and Sense will ever 100% agree on instantaneous readings for a variety of reasons:

  • Solar fluctuates considerably and getting simultaneous screenshots is difficult.

  • IoTaWatt damps the status display so it is not an exact measure at any point in time.

  • Sense probably damps as well and there could also be a latency between the upload to the cloud and query by the app.

  • IoTaWatt, as configured with a 120V AC adapter, is subject to error from voltage swings between the two phases caused by load and the resistance of your neutral cable. (That can be eliminated by using a 240V VT). Sense may be using the 240V measurement, or may be using one of the 120V measurements.

These problems make the instantaneous reading hard to compare, but the energy readings should be very accurate, at least with IoTaWatt. You can track the IoTaWatt kWh over a couple of days vs the meter. You can also track the IoTaWatt solar generation kWh against the inverter display. The same should be possible with the Sense and will show best where these errors lie.

UPDATE: Original post had Type-1 and Type-2 solar incorrectly labelled. This has been corrected. Note also that the post had assumed the wrong type and that is still the case. Subsequent explanation by @RedTechie has clarified the type.

1 Like

Sorry, your post came in while I was working on the above response. It explains a lot.

Girst, what I think has happened it that you have crossed the phase A and phase B mains CTs. I notice in the configuration that one has the reverse box checked, so switching the two would cause both to reverse, which would be easy to do with extensions etc.

So this will resolve the comments above about developing consumption and negative mains during export, but please do look into the issue of the two inverter CTs and that they disagree. (The plot actually shows they are closer than the status display indicates).

Wow lots to unpack here! Let me do my best!

I’m not seeing a negative input except the TeslaSubPanel_PhaseA.

I only was seeing negative values on my Utility lines when solar wasn’t generating (so last night). I never have seen my utility go into the negatives when solar wasn’t generating before which tipped me off to an issue. When solar is generating I am seeing both solar and utility in the positives (see the graph+ image I posted above)

You seem to have CTs on both legs of the solar inverter output. Most inverters are two-wire and output the same on both legs. The IoTaWatt is reporting significantly different power on the two legs, which should not be the case. I believe the Sense only uses one CT for the solar. Is that true in your case?

I don’t have an inverter, my system is micro-inverters behind each panel on the roof. They all come into a separate breaker box and the mains from that breaker box are crimped onto the mains of my utility (before my main 200amp house breaker)

A 20,000Watt solar array would put out more than 80Amps on each leg. I’m wondering how that is fed into your system. There are two different approaches:

Yeah, we have a huge 78 panel system capable of 24kw of generating power. The fused discount is 125amp. We are using Type 1 in your example.

Moving on from that, I think the Sense and IoTaWatt agree within 1% of the Mains power.

Don’t get me wrong, I’m not trying to compare the two in superior performance. I’m just using sense as a rough guide to make sure the data I’m getting from IotaWatt is correct. I just included a screen cap of it to help give you some context when debugging the situation. Nothing more :slight_smile:

I suggest that you should check “reverse” for UtilityMain_PhaseA and uncheck "reverse for UtilityMain_PhaseB.

Bingo!!! So far that looks much better now! We’ll see what it looks like tonight when solar isn’t generating but right now with solar generating its now showing negative values for utility! Not sure why I had to do that… And I swear I tried that last night, but I’ll take it if it keeps working tonight as well!

IoTaWatt, as configured with a 120V AC adapter, is subject to error from voltage swings between the two phases caused by load and the resistance of your neutral cable. (That can be eliminated by using a 240V VT). Sense may be using the 240V measurement, or may be using one of the 120V measurements.

So this was going to be another question I had (unrelated). I’m aware Sense reads the voltage from each phase directly. I actually still use my hidden API to get that data for Grafana. My question to you is I seem to calibrate the voltage on the Iotawatt’s and get it dialed in precisely! But then a few hours later its wildly off by 5-6 volts. Is there a trick I’m missing here?

UPDATE: Solar type has been corrected.

We’re crossing in the night here.

First, I got the two PV types backwards. The OEM folks defined these terms and so I’m trying to be consistent. I’ll correct the above post to reflect the correct terms.

Since you have type-1 solar your mains should never go negative. There’s nothing downstream to produce power and cause it to flow the other way. But your post sheds more light on this.

My second post above suggested that the reversal of the mains plot was caused by crossing the mains CTs. I still think that’s the case. The solution, in that context, was to switch the inputs used for the CTs. What you did in clicking the reverse boxes will also reverse them, but they will still be plugged into the wrong inputs (if they were correct before). you need to reverse back and swap the inputs to make the mains positive again and plugged into the correct inputs.

With your type-1 solar, the math is:

Consumption = sum of the mains.

Generation = doubled value from one Solar CTs.

Net Utility Power = Consumption - Generation

Current Export = (Consumption - Generation) min 0 abs

Current Import = (Consumption - Generation) max 0

Note to anyone reading this for advice, this is a type-1 solar install. Most US solar is type -2, which is configured differently.

UPDATE: I will soon create a page in the docs to clarify all of this.

Let me bring it all together :slight_smile:

Sorry, your post came in while I was working on the above response. It explains a lot.

First, what I think has happened it that you have crossed the phase A and phase B mains CTs. I notice in the configuration that one has the reverse box checked, so switching the two would cause both to reverse, which would be easy to do with extensions etc.

So this will resolve the comments above about developing consumption and negative mains during export, but please do look into the issue of the two inverter CTs and that they disagree. (The plot actually shows they are closer than the status display indicates).

Yes, swapping the reverse checkbox on both UtilityMain inputs fixed the issue as of right now. And I’m now getting a correct consumption readings based on my control (Sense Meter).

Blue circle is the time frame I reversed the reverse checkboxes in Iotawatt. Light red line is Sense consumption, dark red line is IotaWatt Consumption.

Since you have type-1 solar your mains should never go negative. There’s nothing downstream to produce power and cause it to flow the other way.

Also this may help clear up any confusion. The order of CT clamps for Solar and Utility in my setup look like this (for the most part):


Keep in mind this is an old picture and doesn’t have the IotaWatt CTs pictured as this was from January. But the IotaWatt CT for UtilityMain are where the Sense CTs are at the top of the picture. The IotaWatt CT’s for Solar are where the Sense CTs are at the bottom of this picture. Also the additional breaker you see in the top left box where the Utility CT clamps are is for the Tesla charger. They wanted to wire it in line side of the whole house generator so it didn’t overload the generator.

I’ll be posting in the case study forum when my IotaWatt setup is complete. I think you may like it :slight_smile:

But back to the point, I’m a bit surprised by the fact that you said it should never go negative. As wouldn’t it go negative if your producing more than your using?

My second post above suggested that the reversal of the mains plot was caused by crossing the mains CTs. I still think that’s the case. The solution, in that context, was to switch the inputs used for the CTs. What you did in clicking the reverse boxes will also reverse them, but they will still be plugged into the wrong inputs (if they were correct before). you need to reverse back and swap the inputs to make the mains positive again and plugged into the correct inputs.

So are you saying I should unplug what’s plugged into Input 1 and plug it into Input 2? Then plug what was in Input 2 to Input 1? Or are you saying I should twist around (reverse) the CT clamp in the panel for the two Utility clamps?

I’m sure i didn’t mix up the inputs as I made sure I labeled everything. I also followed the scheme of following the main red cables in the panel as PhaseB and the main Black cables as PhaseA. So I’m at a loss as to why it would be mixed up. I do remember the Solar folks having some weird issues with phases when they were trying to get their CT clamps working (the black CT clamps you see in the picture above is for Utility monitoring in the Solar app, the solar generation CT’s aren’t pictured here, there in another panel to the right). So I’m not sure if I have a systemic issue with phase polarity here.

But I guess more importantly, do you think I need to swap out the 24 TRRS (4 conductor) 3.5mm cables for TRS (3 conductor) 3.5mm cables based on this information? Will the continued use of the TRRS cables damage the IotaWatt or the CTs?

Short answer is that I don’t see evidence they are causing this problem.

Based on this

That’s clearly true, but missing was the picture that shows the mains CTs are before where the solar ties in. It’s a Type-2. The solar can drive the mains negative.

So as it sits, it looks OK.

Same here. I don’t have a sense, but it looks like a quality product and notwithstanding the shortcomings people cite concerning individual device measurement, I think it should be doing a good job of the mains and solar where it has dedicated CTs. I think IoTaWatt and the Sense should pretty much agree, especially on kWh over time.

But IoTaWatt appears to be consistently showing higher solar generation, and I think that’s a problem. Also, the mains are farther apart than I would like. Looking at the status display in one of the above posts, the IoTaWatt is showing 126.9 Volts. That’s pretty high. I’m wondering if there is a calibration issue. Have you checked the calibration with a meter, or does sense give you voltage that you can use to see if it’s in the ballpark?

127 Volts is 5.8% higher than 120V. IoTaWatt solar was about 3.5% higher than sense. So it could easily be just a voltage calibration issue.

Also, the mains are farther apart than I would like.

Yeah, this isn’t an IotaWatt problem. I have a 42U rack in the basement with a lot of servers/network equipment. I sadly made the mistake of wiring it up as a 30a 120v circuit vs. a 240v circuit. The rack pulls about ~1kw of power all on one phase and thus throws off my two phases quite a lot in terms of power balance. Which is why I want to monitor each phase independently and not just double up one phase as I will be missing a lot of data.

Looking at the status display in one of the above posts, the IoTaWatt is showing 126.9 Volts. That’s pretty high. I’m wondering if there is a calibration issue. Have you checked the calibration with a meter, or does sense give you voltage that you can use to see if it’s in the ballpark?

So this is in regards to my question that may have been lost in the shuffle:

So this was going to be another question I had (unrelated). I’m aware Sense reads the voltage from each phase directly. I actually still use my hidden API to get that data for Grafana. My question to you is I seem to calibrate the voltage on the Iotawatt’s and get it dialed in precisely! But then a few hours later its wildly off by 5-6 volts. Is there a trick I’m missing here?

I use a kill-a-watt meter to calibrate the VT inputs but every time I calibrate and get them to exactly what the voltage reads on the kill-a-watt meter, a few hours later they are completely off again. For example I calibrated last night and they are already considerably off again (right now I am off by 6-7 volts). Not sure how to get them to lock in. It’s probably my biggest issue with the IotaWatt compared to the Sense is lack of detailed voltage per phase which can throw off all the data it collects. :frowning: Which is why I still log individual phase voltages from Sense in Grafana as they are direct readings.

Sorry, I missed that part about voltage drift. IoTaWatt is not drifting 5-6 Volts. Lets try top resolve this.

Are you sure that when you are calibrating, the IoTaWatt VT is plugged into the same phase as the Kill-a-Watt? With the imbalance you described, that sounds like a possibility.

Are there any large loads on the VT circuit?

Can you plot the sense voltages with the IoTaWatt voltage?

Given the nature of this setup, I’d like to propose that you add another VT for the other phase. Your well pump should be two-wire so you can use just one CT for that. If you can free up input 13 or 14 I can send you another VT. With a second VT, you can get exact voltages for each of the phases on the mains, solar, and Tesla. It will also require that you have an outlet for each phase.

Are you sure that when you are calibrating, the IoTaWatt VT is plugged into the same phase as the Kill-a-Watt? With the imbalance you described, that sounds like a possibility.

Yup, the kill-a-watt is inline with the cable to the power strip that is powering both VTs. I have two IotaWatts so far (named M01 and M02). Will probably be getting two more for the basement sub-panels (B01 and B02) at some point soon.

Are there any large loads on the VT circuit?

Not really, There is a garage refrigerator and a few other odds and ends. The IotaWatts are currently plugged into a GFCI outlet in the garage next to the main panel. But funny story is that outlet isn’t wired into the main panel in the garage. It’s wired into the sub-panel in the basement as they needed more room in our main panel. Our main panel is 100% full.

Can you plot the sense voltages with the IoTaWatt voltage?

Great idea!

Today so far (I did re-calibrate right after my last post which brought the vales back into range):

Last 2 Days:

Last 7 Days:

Last 30 Days:

Closer look at the incident that seemed to have cause the voltage to get out of range around 12:22am last night. Not really sure what would have caused this:

After graphing this; the data shows that the voltages between the Iotawatt’s and the Sense readings seem to be closer than I thought. But something keeps whacking them out of range every so often.

Given the nature of this setup, I’d like to propose that you add another VT for the other phase. Your well pump should be two-wire so you can use just one CT for that. If you can free up input 13 or 14 I can send you another VT. With a second VT, you can get exact voltages for each of the phases on the mains, solar, and Tesla. It will also require that you have an outlet for each phase.

I appreciate it! But I currently have two Iotawatts and VT’s. I would have loved to have one plugged into a phase A and another plugged into a phase B. But the problem is we have no more circuits in the panel available and only one outlet next to the panel (you can just barely see it in the picture in the previous post). I have seen some folks on these forums use hard wired in transformers which I could do if I share a double pole circuit with two transformers one going to each IotaWatt to log Phase A and Phase B. Maybe I’ll do that in the future…

The data from the Sense shows that your two phases don’t really diverge much, so I don’t think another VT is going to resolve this.

Looking at your configuration, you have the VT configured as a TDC DA-10-09. I haven’t shipped those for a long time, so I’m wondering if you have a TDC DA-10-09-E6. See this in the documentation. Welcome to my world, trying to get commodity components from China. I tried to get them to use a different model number, but they will not. I believe it has to do with the UL listing and the cost to list a new model.

In any event, what I think is happening is that you are calibrating to the newer E6 calibration and then it’s reverting to the older model calibration when it restarts or some other event. I’ll look into that, but if you have the E6, you can just change it in setup. The standard calibration for that model should be fine, and it should not change.

I’m pretty sure this is why the IoTaWatt is reading higher than the Sense.

Ah! I totally missed that in the documentation. I noticed the default model number that came configured in the IotaWatt was different than the model on the label of the VT. Now I know why! Thanks for pointing me to those docs. I updated the VTs back to the E6 model. I will double check the calibration also a little later to make sure I didn’t “de-calibrate” when calibrating against the wrong model VT.

1 Like

So I think I found another reason my voltage was so out of range. I should have just used the Sense Meter to calibrate the Iotawatts. As it looks like my Kill-a-watt is not calibrated (go figure). The Kill-a-watt is constantly reading a higher voltage than everything else. Check out the pics below:


In this picture of the Sense app. You can see how much is loaded up on the one phase. :frowning: Which causes the voltage to dip a little bit more than usual on that phase. The Iotawatt’s are plugged into the 117.4 volt phase and they are in lock step with that value now.

Also after changing the Iotawatt VT model and re-calibrating, the voltages are looking super tight now! :slight_smile:

At some point what I may do is get Iotawatt-M02 (or one of my future basement Iotawatts) on the other voltage phase so I can get that voltage reading from Iotawatt as well, eliminating the need for the undocumented Sense APIs.

Thanks for all the help @overeasy it’s appreciated! I’ll be posting the finished project in the case studies category soon, after a few finishing touches to the graphs and physical install! :slight_smile:

1 Like