Two IotaWatts, two VT's, split phase

I am putting in two IotaWatt’s, on opposite sides of my US Split phase 200A/240V panel.

Because it is easy, I think I will run two VT’s one on each leg, split both, and feed both to both IotaWatt’s. At worst I’ll see nicely balanced voltage all the time; at best I might see something interesting.

Let’s call the VT’s L1 and L2 for each leg, and the IotaWatt’s Left and Right.

Left and Right’s USB power adapter will be plugged into different legs for each Iotawatt, but I assume that does not matter (it’s all DC, right?).

One VT will go on the 9V Ref input, the other on a CT input on each IotaWatt. Should I use L1 on both left and right IotaWatt’s Ref input (with L2 on another input).

Or should L1 go on the Left and L2 on the Right (and thuse L2 and L1 on the Ct input on each)?

Or does it matter?

Is there guidance somewhere how to make use of it now that I have a VT for each leg? In browsing the input definitions I do not see a place to select the reference source.

Yes, this is unlikely to yield anything terribly worthwhile, but gaining some education is always interesting, and it can be another data point when someone asks “is it worth it”.


It’s arbitrary, but I would recommend you do it the same on each unit to avoid confusion. For instance Input 0 L1 and input 14 L2.

I must caution you to insure polarity is consistent. Splitters are cheap from any shop that supplies the CCTV industry.

When you configure more than one VT, a menu item will be added to the CT configuration to select the voltage source. It will initially default to the channel 0 VT but the dropdown will include the other configured VTs.

Aside: I’m delighted that USPS ran early, and my units arrived today, so am setting up now.

I must caution you to insure polarity is consistent. Splitters are cheap from any shop that supplies the CCTV industry.

I’m feeling a bit dense now that I see these - the inputs are headset type plugs (3 conductor type). The reference is a typical 9V two conductor. I eee lots of splitters but none to convert. Are you supposed to wire your own adapter? If so, which pins on the 3 wire?

Or… is the 3phase reference input (one only) usable as opposed to input 14?

You will see that the three-phase inputs are labelled 13 and 14. When you use one of those for a VT, do not use the corresponding 3.5mm CT input. Internally, it will automatically switch to whichever has a plug inserted.

The splitters for 5.5mm x 2.1mm barrel jacks are commonly sold by the CCTV industry as DC power splitters, but they work just as well for AC. Just google “5.5mm splitter”.

Ah… I didn’t see that aspect. Perfect. I had an extra VRef transfomer (ordering error, mea culpa, no issue) so I did that for now to test on one side, and it worked perfectly. Now I need to go through and make sure I know which leg is where, maybe flip a few CT’s (I started making them match the IotaWatt’s reference, but I really should have done it physically consistently, then later flip the reference).

The difference is huge, about half a volt (uncalibrated). :upside_down_face:

Now to get some splitters and next figure out Influxdb and link to Home Assistant and … I guess now I have to actually DO something with all this info. :thinking:

Extremely straightforward to install, I just made it more complex with two VRefs. Thank you for the quick response.

OK, recognizing that no one need do this generally, and that it really doesn’t move the needle much, I do have one minor thought if you are ever looking for something to program…

For pure 240V loads ideally the Vref should be the sum of these two references without doubling, not either/or, if one is using one CT. Obviously you can just do another CT.

Please give this all the priority in your programming it deserves. :smirk:

That’s a more complicated, yet more useful approach than you may think.

On the complicated side, it’s not a matter of just adding up two voltages. The IoTaWatt is sampling voltage about 640 times over a cycle and reporting the rms value of those readings. Reading another voltage involves reading another VT, which would complicate the sampling at best and slow it down by 1/3 at worse. But I have given that thought because…

On the useful side, it has practical application in three-phase systems. When you place a load across two phases of a three phase system, the resultant voltage signal is sqrt 3 greater and not twice as in a split-phase system. Moreover the phase is significantly different where it is the same in a split-phase system. So using the sum of the voltage on the two phases would give the correct result in both cases.

At the risk of TMI, the way sampling works in the IoTaWatt is to sample the voltage and current alternately. Using the raw samples would result in a sampling phase shift of about 360/640° or 0.56°. So the IoTaWatt develops sample pairs by averaging the voltage readings that bracket each current sample. That effectively synchronizes the voltage sample with the current sample. Averaging is linear interpolation, which works fine on the voltage sample at these high sample rates.

So the idea I have been toying with is to alternately sample voltage from the two VTs and use the sum rather than the average in these cases. As it happens, the beauty of that approach would be that the technique would do exactly what you are suggesting for split phase with two VTs. Of course it’s a little more complicated than even alternating the VTs. The zero crossing detection would need to be able to handle that as well.

If you ever see this, it will be in the ESP32 version. In the meantime, I stick to my recommendation to use a single 240V VT for simplicity with optimum results in a split phase system.

Never TMI, always interested, thank you. So if (emphasis on “if”) one had legs with substantially different voltages (which I do not), two CT’s and two VRef’s for now is more accurate.

I was actually wondering, staring at the status screen, about the sampling rate/technique. If I really get bored I’ll start looking at the code as light reading.

My guess is that in my case putting two CT’s on a pure 240V load probably has more measurement error than the difference the voltage will make.